Subtitle becomes text.
Then have text content/decoder which can be used-as subtitle or ccap.

Doubling-up of subtitle/ccap starts in player: that emits subtitle or closedcaption, writer writes both.
libdcp gets closedcaptionasset with some common ground with subtitleasset, maybe
add_text_asset() which is called for subtitle + ccap by ReelWriter::create_reel()

CPLs can have CCAP and subtitle; actual files appear to be the same as subs.

Maybe we just want a subtitle type with the emissions e.g. instead of emit(sub) do emit(TEXT_SUBTITLE, sub) or emit(TEXT_CCAP, sub).

Try "bitmap" for bitmapped text and "plain" for actual text?

Should we try to preserve project back-compatibility here (i.e. do we change XML tag names...)
on the one hand, it's just names, on the other hand...

Content needs a list<TextContent> since DCP content may have multiple bits.